sacdallago/bio_embeddings

Supports multiple state-of-the-art language models (SeqVec, ProtTrans, ESM, UniRep) with a unified Python interface, enabling transfer learning for downstream tasks like structure/function prediction. The pipeline generates per-amino-acid and per-sequence embeddings, applies dimensionality reduction (UMAP/t-SNE), and enables both supervised and unsupervised annotation extraction. Includes a distributed webserver API and Docker deployment for reproducible, scalable workflows with GPU acceleration and automatic out-of-memory handling.

507 stars. No commits in the last 6 months.